CN1404249A - Double-weighing parallel interference-counteracting algorithm - Google Patents

Abstract

The present invention provides a cancellation algorithm of bilayer weighted parallel interference based on the studing of cancellation algorithm of partial parallel interference and based on the weighted parallel interference cancellation algorithm of the Bayes theorem. The algorithm not only has the advantage of weighted algorithm based on the Bayes, theorem which can achieve the minimum cost in bit grade judgement, but also can make up the deviation to the client signal estimation on the statistic means at the advantage possessing in partial weighted algorithm. At the same time, comparing with the cancellation algorithm of weighted parallel interference based on the Bayes theorem, the algorithm has increased the gain a lot at the low signal to noise ratio without mush increasing amount of calculations.

Description

Double-weighing parallel interference cancellation algorithm

Technical field

The present invention relates to the multiuser detection of base station in the CDMA mobile communication system, relate in particular to parallel interference cancellation algorithm in the cdma system.

Technical background

Cdma system has become the developing direction of 3G (Third Generation) Moblie because of advantages such as its high power capacity, high quality-of-service, good confidentiality.Multiple access disturbs (Multiple Access Interference) to limit the raising of cdma system capacity and performance.A user disturbs very little usually to the multiple access that other users produce.But when number of users increased, to some users, the multiple access that other users produce it disturbed summation just very big.When having near-far interference, certain user's signal amplitude is bigger, and the multiple access that this user produces the weak signal user disturbs just bigger, and weak signal is submerged among the strong signal.Single receiver user can't be eliminated multiple access and disturb the influence that subscriber signal is detected, and the detection performance of receiver reduces under number of users increase and near-far interference situation.The receiver that designs anti-multiple access interference is the key of performance cdma system high power capacity, high quality-of-service advantage.

Multiuser detection is to overcome the influence that multiple access disturbs, and improves a kind of Enhanced Technology of cdma system capacity.It can make full use of a plurality of users' information, and a plurality of subscriber signals are carried out joint-detection, disturbs the receiver Effect on Performance thereby reduce multiple access as much as possible, improves the capacity of system.

The multi-user detector that Verdu proposed based on maximum a posteriori probability in 1986, i.e. maximum likelihood sequence detector.Though this detector is an optimal detector, this detector complexity height, and need the estimation of amplitude and phase information to received signal.This makes that maximum likelihood sequence detector is difficult to use.Therefore, must study the multiuser detection algorithm of suboptimum.

The multiuser detection algorithm of suboptimum roughly is divided into two classes: linearity test algorithm and interference cancellation algorithm.The linearity test algorithm carries out linear transformation to the soft output of single user detector, produces one group of new output that can improve performance.This class algorithm mainly comprises: decorrelation detector (Decorrelating Detector), minimum mean square error detector (MinimumMean Square Error Detector) and polynomial expansion method (Polynomial Expansion Detector) etc.The linearity test algorithm performance is better, but calculates very complicated.The interference cancellation algorithm is considered as useful signal with the signal of desired user, and other users' signal is considered as interference signal; Eliminate other users' interference earlier from received signal, obtain the signal of desired user, the signal to desired user detects then, thereby improves the performance of system.The interference cancellation algorithm can be divided into: successive interference cancellation (Serial Interference Cancellation) and parallel interference cancellation (ParallelInterference Cancellation).The successive interference cancellation algorithm sorts to subscriber signal according to the power descending.At first prominent user is adjudicated detection, this subscriber signal of regenerating then removes this user's signal from received signal, makes other user's detection not be subjected to the interference of this subscriber signal.Next the inferior maximum subscriber signal of power is detected, the user's of regeneration and elimination power time maximum signal makes remaining user's detection no longer be subjected to power time maximum user's interference.In order, from received signal, remove other users' interference again.The performance of this method is better than single user detector; But shortcoming is: (1) time-delay is bigger; (2) need carry out the power ordering, amount of calculation is bigger; (3) initialize signal is estimated sensitivity.The parallel interference cancellation algorithm is eliminated every other user's signal interference concurrently for each user from received signal.This algorithm performance is better than single user detector, and it is little to have a time-delay, and the advantage that computational complexity is little is the at present most possible algorithm of realizing.

The parallel interference cancellation algorithm improves bigger with respect to the single user detector performance under high s/n ratio; But under low signal-to-noise ratio, this algorithm reduces with respect to the amplitude that the single user detector performance improves.In cdma system, power control can remedy the fading characteristic of channel to a certain extent, system is being worked, to improve the capacity of system as much as possible than under the low signal-to-noise ratio.Therefore, how to improve significant than the performance of parallel interference cancellation algorithm under the low signal-to-noise ratio.

Part parallel interference cancellation algorithm can effectively improve the performance of parallel interference cancellation algorithm.Different with traditional parallel interference cancellation algorithm is: traditional parallel interference cancellation algorithm is fully eliminated the multiple access interference that desired user is subjected to from received signal; And part parallel interference cancellation algorithm is provided with weights for every grade of interference cancellation, and the multiple access interference that desired user is subjected to is weighted, and in the interference cancellation process, just partly eliminates multiple access and disturbs.R.Michael Buehrer and Steven P.Nicoloso are at communication newspaper (the IEEETransactions on Communications of the fifth phase in 1999 Electrical and Electronic engineering association publication, PP.658-661, Vol.47, No.5,1999) delivered " note of " the part parallel interference cancellation that is used for CDMA " " (Comments on " Partial Parallel InterferenceCancellation for CDMA ") on.This piece paper is analyzed theoretically and obtained: traditional parallel interference cancellation algorithm is fully eliminated the multiple access interference that desired user is subjected under the Gaussian channel from received signal, and the estimation to the desired user signal is that inclined to one side estimation is arranged in this case; Part parallel interference cancellation algorithm is just partly eliminated multiple access and is disturbed, and can correct the deviation that the desired user signal is estimated, makes court verdict more reliable.Than under the low signal-to-noise ratio, the performance of part parallel interference cancellation algorithm obviously is better than traditional parallel interference cancellation algorithm.

Weighing parallel interference cancellation algorithm based on bayesian criterion is open by U.S. Pat 5418814, and it also is a kind of weighting algorithm.It is different with the weighting base reason of part parallel interference cancellation algorithm, and it is based on the bit-level weighting algorithm of the average minimum of judgement cost.This algorithm is provided with the cost function of judgement, with the judgement cost the average minimum as criterion, determine each bit decision result's coefficient of reliability, and the signal of this bit regeneration is carried out the bit-level weighting with this coefficient, in the elimination that multiple access disturbs, just partly eliminate the interference of this bit generation of this user like this.The performance of this algorithm has raising with respect to traditional parallel interference cancellation algorithm, and especially under the low signal-to-noise ratio situation, performance improves more obvious.

Below, this paper introduces traditional parallel interference cancellation algorithm, part parallel interference cancellation algorithm and in detail based on the weighing parallel interference cancellation algorithm of bayesian criterion.

1 traditional parallel interference cancellation algorithm

The structure of parallel interference cancellation receiver as shown in Figure 1.The internal structure of PIC structure 1 among Fig. 1 and afterbody PIC structure 2 respectively as shown in Figures 2 and 3.First order PIC structure is handled the input signal of the baseband signal of received signal as each user, and each user's who obtains output signal is the input signal of each user in the next stage PIC structure; Second level PIC structure is handled each user's input signal, and each user's who obtains output signal is the input signal of each user in the next stage PIC structure; Handle so step by step, afterbody PIC structure is handled each user's input signal, and each user's who obtains output signal is the final result of multistage PIC structure.

Under the fading channel environment, the baseband signal of received signal can be expressed as: $r\left(t\right)=\underset{i=1}{\overset{K}{\mathrm{\Σ}}}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{a}_{\mathrm{il}}{S}_i(t-{\mathrm{\τ}}_{\mathrm{il}})+Z\left(t\right)$ $=\underset{i=1}{\overset{K}{\mathrm{\Σ}}}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{a}_{\mathrm{il}}\sqrt{P_i}{b}_i(t-{\mathrm{\τ}}_{\mathrm{il}})c_i(t-{\mathrm{\τ}}_{\mathrm{il}})+Z\left(t\right)---\left(1\right)$

Wherein, the baseband signal of r (t) expression received signal; a _IlThe channel fading value of representing i user l footpath, L is the footpath number; τ _IlThe time delay of representing i user l footpath; S _i(t) the transmission signal of expression user i, K represents total number of users; P _iThe power of expression user i; b _i(t) bit stream of expression user i, $b_i\left(t\right)=\underset{m=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{a}_i^{\left(m\right)}p(t-m{T}_b),a_i^{\left(m\right)}$ M bit representing i user, p (t) indication cycle is T _bSignal pulse, under the situation that does not hinder algorithm derivation conclusion, establish p (t) and be rectangular pulse (when t ∈ [0, T _b] time, p (t)=1; When The time, p (t)=0); c _i(t) spreading code of expression user i; Z (t) represents interchannel noise.

In k level PIC structure, the input signal of the RAKE receiver 3 of user i is

When k=1 $r_i^{\left(l\right)}\left(t\right)=r\left(t\right)$ 。RAKE receiver is right Carry out multi path despreading, and carry out channel estimating, carry out multipath then and merge by the despreading result.The RAKE receiver of user i to the despreading result in l footpath is:

Wherein, l=1 ..., L.

The employing high specific merges, and the multipath amalgamation result that obtains RAKE receiver is: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}---\left(3\right)$

Wherein, $y_i^{\left(m\right)\left(k\right)}=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}^{*}{y}_i^{\left(m\right)\left(k\right)}\left(l\right)---\left(4\right)$

A _IlBe Estimated value, a _IlThe channel fading value of representing i user l footpath, P _iThe power of expression user i.The multipath amalgamation result of 4 pairs of RAKE receiver of hard decision device carries out hard decision, just obtains the court verdict of k level PIC algorithm.When k=1, this court verdict is exactly the output of single user detector.Court verdict to m bit of i user is: ${\hat{a}}_i^{\left(m\right)\left(k\right)}=\mathrm{sgn}\left\{Y_i^{\left(m\right)\left(k\right)}\right\}---\left(5\right)$

Court verdict is further processed, comprise with signal regenerator 5 carry out signal regeneration, with the multiple access estimation of disturbing and estimation and interference cancellation that interference cancellation device 6 carries out the multiple access interference, just obtain the output signal of user i in the k level PIC structure.This signal is exactly the input signal of the RAKE receiver of user i in (k+1) level PIC structure.In the k level PIC structure, the regenerated signal of user i can be expressed as: $g_i^{\left(k\right)}\left(t\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{\hat{a}}_i^{\left(n\right)\left(k\right)}p(t-n{T}_b-{\mathrm{\τ}}_{\mathrm{il}})c_i(t-{\mathrm{\τ}}_{\mathrm{il}})---\left(6\right)$

By other (K-1) individual users' regenerated signal, can obtain the estimation of the MAI of user i: ${\hat{I}}_i^{\left(k\right)}=\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{g}_j^{\left(k\right)}\left(t\right)$ $=\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{jl}}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{\hat{a}}_j^{\left(n\right)\left(k\right)}p(t-n{T}_b-{\mathrm{\τ}}_{\mathrm{jl}})c_j(t-{\mathrm{\τ}}_{\mathrm{jl}})---\left(7\right)$

To i user, from the baseband signal of received signal, eliminate the MAI that other users produce according to following formula: $r_i^{(k+1)}\left(t\right)=r\left(t\right)-{\hat{I}}_i^{\left(k\right)}---\left(8\right)$

It is the output signal of user i in the k level PIC structure.This signal is the input signal of the RAKE receiver of user i in (k+1) level PIC structure.

As shown in Figure 3, in the afterbody PIC structure of S level PIC algorithm, 3 pairs of input signals of the RAKE receiver of user i Carry out multi path despreading, channel estimating, multipath merging, the soft output that multipath merges the user i that obtains is exactly the final result of user i in the S level PIC algorithm.This result is fed to the decoder of user i and deciphers.Afterbody PIC structure does not comprise estimation and the interference cancellation device that signal regeneration and multiple access disturb.

The performance of PIC algorithm is better than single user detector: when signal to noise ratio was higher, this algorithm had bigger gain with respect to single user detector; When signal to noise ratio was low, this algorithm reduced with respect to the gain of single user detector.

2 part parallel interference cancellation algorithms

Obtain in judgement Under the correct condition, eliminate MAI, can improve the performance of system according to formula (8).But, when When incorrect, have error message among the MAI that estimates to obtain according to formula (7), in the interference cancellation process, not only can not remove corresponding M AI item, increased interference on the contrary, this must make the court verdict error rate of PIC increase.At this situation, part PIC algorithm is weighted the signal of regeneration, carries out interference cancellation then.The structure of the structure of part PIC algorithm and traditional PIC algorithm is identical, as Fig. 1, Fig. 2 and shown in Figure 3.Just in part PIC algorithm, the concrete formula of interference cancellation is as follows: $r_i^{(k+1)}\left(t\right)=r\left(t\right)-p^{\left(k\right)}{\hat{I}}_i^{\left(k\right)}---\left(9\right)$

Wherein, p ^(k)Be the weights of k level PIC algorithm; p ⁽¹⁾＜p ⁽²⁾...＜p ^(S)S is the progression of PIC.

In the first order of part PIC algorithm, the court verdict accuracy rate is lower, and is therefore also unreliable to the estimated result of MAI, so can weighting value p ⁽¹⁾Less, partly eliminate MAI, so both can eliminate a part and disturb, can reduce of the influence of wrong court verdict again to algorithm performance; Along with the increase of PIC progression, the court verdict accuracy rate that the PIC algorithm obtains improves, and is reliable to the estimation of MAI, so can make weights p ^(k)Increase with k progressively increases, and strengthens the elimination dynamics to MAI, to improve the performance of PIC algorithm.At present, have only qualitative conclusions about choosing of weights: along with the increase of PIC algorithm progression, weights p ^(k)Increase.Communication newspaper (IEEE Transactions on Communications, PP.658-661, Vol.47, No.5,1999) delivered " note of " the part parallel interference cancellation that is used for CDMA " " (Comments on " Partial ParallelInterference Cancellation for CDMA ") on.This piece paper is analyzed theoretically and obtained: traditional parallel interference cancellation algorithm is fully eliminated the multiple access interference that desired user is subjected under the Gaussian channel from received signal, and the estimation to the desired user signal is that inclined to one side estimation is arranged in this case; Part parallel interference cancellation algorithm is just partly eliminated multiple access and is disturbed, and can correct the deviation that the desired user signal is estimated, makes court verdict more reliable.

3. based on the weighing parallel interference cancellation algorithm of bayesian criterion

Propose a kind of weighting PIC algorithm based on bayesian criterion in the U.S. Pat 5418814, the multilevel hierarchy of this algorithm is with the traditional PIC algorithm, as shown in Figure 1.The PIC structure of this algorithm is different from the PIC structure of traditional PIC algorithm, and concrete structure is seen Fig. 4.The afterbody PIC structure of this algorithm is with the traditional PIC algorithm, as shown in Figure 3.This algorithm principle is as follows:

By formula (2～4), the multipath amalgamation result of the Rake receiver of user i can further be expressed as in the k level PIC structure: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}={\mathrm{\μ}}_i{a}_i^{\left(m\right)}+n_i---\left(10\right)$

n _iBe white Gaussian noise, Normal Distribution Be m the bit of user i, be worth and be+1 or-1.μ _iBe the real number relevant with channel fading.

By (10) formula, can obtain: when $a_i^{\left(m\right)}=1$ The time, Normal Distribution When $a_i^{\left(m\right)}=-1$ The time, Normal Distribution

If court verdict ${\hat{a}}_i^{\left(m\right)\left(k\right)}=\mathrm{sgn}\left\{Y_i^{\left(m\right)\left(k\right)}\right\}$ Reliability be , the cost function that judgement is set is: $C={\mathrm{\μ}}_i^2{(a_i^{\left(m\right)}-f_i^{\left(m\right)\left(k\right)}{\hat{a}}_i^{\left(m\right)\left(k\right)})}^2---\left(11\right)$

Can obtain adjudicating the average of cost according to Bayes rule: $E\left(C\right)={\mathrm{\μ}}_i^2\{{(1-f_i^{\left(m\right)\left(k\right)})}^2(1-P_e)+(1+f_i^{\left(m\right)\left(k\right)}{)}^2{P}_e\}$ $={\mathrm{\μ}}_i^2\{1-2f_i^{\left(m\right)\left(k\right)}(1-2P_e)+{\left(f_i^{\left(m\right)\left(k\right)}\right)}^2\}---\left(12\right)$

Wherein, P _eBe error probability.

To the following formula differentiate, and make that derivative is 0, obtain making the average minimum of judgement cost $f_i^{\left(m\right)\left(k\right)}=\tanh \left\{\frac{{\mathrm{\μ}}_i\left|Y_i^{\left(m\right)\left(k\right)}\right|}{{\mathrm{\σ}}_i^2}\right\}---\left(13\right)$

When signal regeneration, with Be weights, the regenerated signal of each bit be weighted that the signal that weighting is obtained is as the regenerated signal of this bit.Calculate MAI according to formula (7), carry out interference cancellation according to formula (8).Estimation and interference cancellation process that the regeneration of signal, multiple access disturb are seen Fig. 4.Device 7 calculates the coefficient of reliability of hard decision device court verdict among Fig. 4 according to formula (13).The concrete formula of signal regeneration is as follows: $g_i^{\left(k\right)}\left(t\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{f}_i^{\left(n\right)\left(k\right)}{\hat{a}}_i^{\left(n\right)\left(k\right)}p(t-n{T}_b-{\mathrm{\τ}}_{\mathrm{il}})c_i(t-{\mathrm{\τ}}_{\mathrm{il}})---\left(14\right)$

The content of invention

The objective of the invention is: propose a kind of double-weighing parallel interference cancellation algorithm.This algorithm combines with part parallel interference cancellation algorithm with based on the weighing parallel interference cancellation algorithm of bayesian criterion, increase under the little situation at algorithm complex, algorithm performance is improved, and especially under the low signal-to-noise ratio situation, algorithm performance improves bigger.

The present invention is achieved in that double-weighing parallel interference cancellation (double-weighing PIC) algorithm that the anti-multiple access of a kind of CDMA of being used for (code division multiple access) wireless communication system disturbs, this algorithm combines with part parallel interference cancellation algorithm with based on the weighing parallel interference cancellation algorithm of bayesian criterion, and described algorithm comprises the steps:

(a.) in the k level PIC of described double-weighing PIC algorithm structure, with RAKE receiver input signal is carried out multi path despreading, channel estimating, multipath merging, and hard decision is carried out in the output of RAKE receiver with the hard decision device.(2 ~ 5) obtain corresponding hard decision result according to the following equation.

The RAKE receiver of user i to the despreading result in l footpath is:

Wherein, l=1 ... .L, Represent in the k level PIC structure, the input signal of the RAKE receiver of user i, when k=1, $r_i^{\left(l\right)}\left(t\right)=r\left(t\right)$ , the baseband signal of r (t) expression received signal, c _i(t) spreading code of expression user i, τ _IlThe time delay of representing i user l footpath, L is the footpath number, T _bThe cycle of expression signal pulse;

The employing high specific merges, and the multipath amalgamation result that obtains RAKE receiver is:

The employing high specific merges, and the multipath amalgamation result that obtains RAKE receiver is: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}---\left(3\right)$

Wherein, $y_i^{\left(m\right)\left(k\right)}=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}^{*}{y}_i^{\left(m\right)\left(k\right)}\left(l\right)---\left(4\right)$

A _IlBe Estimated value, a _IlThe channel fading value of representing i user l footpath, P _iThe power of expression user i.L is the footpath number.

Court verdict to m bit of i user is: ${\hat{a}}_i^{\left(m\right)\left(k\right)}=\mathrm{sgn}\left\{Y_l^{\left(m\right)\left(k\right)}\right\}---\left(5\right)$

(b.) the every bit decision result's of calculating reliability.

In the k level PIC structure, the court verdict of m the bit of user i is

, formula (3) can further be expressed as: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}={\mathrm{\μ}}_i{a}_i^{\left(m\right)}+n_i---\left(10\right)$

n _iBe white Gaussian noise, Normal Distribution

Being m the bit of user i, being worth and being+1 or-1, is the real number relevant with channel fading,

Coefficient of reliability be calculated as follows: $f_i^{\left(m\right)\left(k\right)}=\tanh \left\{w\frac{{\mathrm{\μ}}_i\left|Y_i^{\left(m\right)\left(k\right)}\right|}{{\mathrm{\σ}}_i^2}\right\}---\left(15\right)$

Promptly be

Coefficient of reliability.W is an arithmetic number, is used for remedying the inaccurate of noise power estimation.The numerical value of w under the different signal to noise ratios can be determined by experiment, when signal to noise ratio is higher, w=1 can be got.

(c.) the bit-level weighting of subscriber signal regeneration.

According to the bit-level weighting regenerated signal of formula (14) calculating user i, its expression formula is: $g_i^{\left(k\right)}\left(t\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{f}_i^{\left(n\right)\left(k\right)}{\hat{a}}_i^{\left(n\right)\left(k\right)}p(t-n{T}_b-{\mathrm{\τ}}_{\mathrm{il}})c_i(t-{\mathrm{\τ}}_{\mathrm{il}})---\left(14\right)$

Wherein, p (t) indication cycle is T _bSignal pulse;

(d.) calculating of multiple access interference. ${\hat{I}}_i^{\left(k\right)}=\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{g}_j^{\left(k\right)}\left(t\right)---\left(16\right)$

(e.) interference cancellation.

If the weights of described double-weighing PIC algorithm k level are p ^(k), according to the following equation the MAI that obtains in the steps d is weighted interference cancellation, obtain the output signal of user i in the k level PIC structure It is the input signal of the RAKE receiver of user i in the next stage PIC structure. $r_i^{(k+1)}\left(t\right)=r\left(t\right)-p^{\left(k\right)}{\hat{I}}_i^{\left(k\right)}---\left(9\right)$

(f.) repeating step (a. ~ e.), carry out the calculating of next stage PIC.For afterbody PIC structure, only carry out step 1 in the calculating that merges of multi path despreading, multipath.That is: according to formula (2) input signal of user i is carried out multi path despreading, carry out multipath according to formula (3), (4) and merge.Multipath is merged the final result of the soft output of the user i that obtains as user i in the multistage PIC structure.In receiver, this result is fed to the decoder of user i and deciphers.

The multilevel hierarchy of double-weighing PIC algorithm is with the traditional PIC algorithm, as shown in Figure 1.The PIC structure of this algorithm is with the weighting PIC algorithm based on bayesian criterion, as shown in Figure 4.Just in double-weighing PIC algorithm, interference cancellation carries out according to formula (9); In the weighting PIC algorithm based on bayesian criterion, interference cancellation carries out according to formula (8).The afterbody PIC structure of this algorithm is with the traditional PIC algorithm, as shown in Figure 3.

The present invention combines with part parallel interference cancellation algorithm with based on the weighing parallel interference cancellation algorithm of bayesian criterion, proposes double-weighing parallel interference cancellation algorithm.This algorithm not only has the advantage based on the weighting algorithm of bayesian criterion, in bit-level judgement cost minimum, and has the advantage of partial weighting algorithm, can remedy the deviation of on the statistical significance subscriber signal being estimated.Simultaneously, compare with weighing parallel interference cancellation algorithm based on bayesian criterion, increase under the little situation in amount of calculation, improved the gain of algorithm when low signal-to-noise ratio greatly, the performance that makes algorithm all improves a lot with respect to the partial weighting algorithm with based on the weighting algorithm of bayesian criterion.

Description of drawings

Fig. 1: the multilevel hierarchy schematic diagram of parallel interference cancellation receiver

Fig. 2: PIC structural representation

Fig. 3: afterbody PIC structural representation

Fig. 4: based on the PIC structural representation of the weighing parallel interference cancellation algorithm of bayesian criterion

Embodiment

Below in conjunction with drawings and Examples the present invention is described in further details.

The multilevel hierarchy of double-weighing PIC algorithm is with the traditional PIC algorithm, as shown in Figure 1.The PIC structure of this algorithm is with the weighting PIC algorithm based on bayesian criterion, as shown in Figure 4.The afterbody PIC structure of this algorithm is with the traditional PIC algorithm, as shown in Figure 3.

As shown in Figure 1, the baseband signal r of received signal (t) enters first order PIC structure 1 among the figure with parallel mode.As shown in Figure 4, and the input signal that is advanced into the PIC structure enter each user's RAKE receiver 3 respectively.RAKE receiver 3 is carried out despreading to input signal earlier, carry out channel estimating by the despreading result then, carrying out multipath at last merges, and give hard decision device 4 and decision reliability calculator 7 simultaneously with the multipath amalgamation result, give decision reliability calculator 7 and signal regenerator 5 simultaneously with channel estimation results.4 pairs of input signals of hard decision device carry out hard decision, and give signal regenerator 5 with the hard decision result.

In the k level of described double-weighing PIC algorithm, (2～5) obtain corresponding hard decision result according to the following equation.

The RAKE receiver of user i to the despreading result in l footpath is:

Wherein, l=1 ... .L,

Represent in the k level PIC structure, the input signal of the RAKE receiver of user i, when k=1, $r_i^{\left(l\right)}\left(t\right)=r\left(t\right)$ , the baseband signal of r (t) expression received signal, c _i(t) spreading code of expression user i, τ _IlThe time delay of representing i user l footpath, L is the footpath number, T _bThe cycle of expression signal pulse;

The employing high specific merges, and the multipath amalgamation result that obtains RAKE receiver is: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}---\left(3\right)$

Wherein, $y_i^{\left(m\right)\left(k\right)}=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}^{*}{y}_i^{\left(m\right)\left(k\right)}\left(l\right)---\left(4\right)$

A _IlBe Estimated value, a _IlThe channel fading value of representing i user l footpath, P _iThe power of expression user i.L is the footpath number;

Court verdict to m bit of i user is: ${\hat{a}}_i^{\left(m\right)\left(k\right)}=\mathrm{sgn}\left\{Y_i^{\left(m\right)\left(k\right)}\right\}---\left(5\right)$

Reliability calculator is calculated the coefficient of reliability of hard decision device court verdicts by two input signals, and gives signal regenerator 5 with coefficient of reliability.In the k level PIC structure, the court verdict of m the bit of user i is

Coefficient of reliability be calculated as follows: $f_i^{\left(m\right)\left(k\right)}=\tanh \left\{w\frac{{\mathrm{\μ}}_i\left|Y_i^{\left(m\right)\left(k\right)}\right|}{{\mathrm{\σ}}_i^2}\right\}---\left(15\right)$

Promptly be

Coefficient of reliability, w is an arithmetic number, is used for remedying inaccurate that noise power estimates, when signal to noise ratio is higher, can get w=1.

Signal regenerator obtains user's regenerated signal by three signals of input, and regenerated signal is sent into the estimation and the interference cancellation device 6 of multiple access interference.The bit-level weighting regenerated signal of subscriber signal is: $g_i^{\left(k\right)}\left(t\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{f}_i^{\left(n\right)\left(k\right)}{\hat{a}}_i^{\left(n\right)\left(k\right)}p(t-n{T}_b-{\mathrm{\τ}}_{\mathrm{il}})c_i(t-{\mathrm{\τ}}_{\mathrm{il}})---\left(14\right)$

As can see from Figure 4: the baseband signal r of received signal (t) also enters estimation and the interference cancellation device 6 that multiple access disturbs.This device estimates that by each user's of parallel input regenerated signal the multiple access that each user is subjected to disturbs, and eliminates signal that multiple access interference that certain user is subjected to the obtains input signal as this user in the output signal of this user in the PIC structure at the corresponding levels, the next stage PIC structure from the baseband signal r (t) of received signal.

In the described double-weighing PIC of the k level algorithm, the multiple access of user i disturbs being estimated as of (MAI): ${\hat{I}}_i^{\left(k\right)}=\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{g}_j^{\left(k\right)}\left(t\right)---\left(16\right)$

If the weights of described double-weighing PIC algorithm k level are p ^(k), according to the following equation the MAI that obtains in the above-mentioned formula is weighted interference cancellation, obtain the output signal of user i in the k level PIC structure

It is the input signal of the RAKE receiver of user i in the next stage PIC structure. $r_i^{(k+1)}\left(t\right)=r\left(t\right)-p^{\left(k\right)}{\hat{I}}_i^{\left(k\right)}---\left(9\right)$

Next stage PIC structure is carried out same processing to the signal of parallel input.Handle so step by step, when handling to the end one-level PIC structure, the signal of parallel input enters each user's RAKE receiver 3 respectively.User's RAKE receiver is carried out despreading, channel estimating and multipath to input signal and is merged, and obtains user's soft output.Each user's soft output is exactly the final result of each user in the multistage PIC structure.In receiver, user's soft output is fed to user's decoder and deciphers.

Claims (1)

1, a kind of double-weighing parallel interference cancellation (double-weighing PIC) algorithm that is used for the anti-multiple access interference of CDMA (code division multiple access) wireless communication system, it is characterized in that, described algorithm combines with part parallel interference cancellation algorithm with based on the weighing parallel interference cancellation algorithm of bayesian criterion, comprises the steps:

(a.) in the k level PIC of described double-weighing PIC algorithm structure, with RAKE receiver input signal is carried out multi path despreading, channel estimating, multipath merging, and hard decision is carried out in the output of RAKE receiver with the hard decision device, (2 ~ 5) obtain corresponding hard decision result according to the following equation

The RAKE receiver of user i to the despreading result in l footpath is:

Wherein, l=1 ..., L, Represent in the k level PIC structure, the input signal of the RAKE receiver of user i, when k=1, $r_i^{\left(l\right)}\left(t\right)=r\left(t\right)$ , the baseband signal of r (t) expression received signal, c _i(t) spreading code of expression user i, τ _IlThe time delay of representing i user l footpath, L is the footpath number, T _bThe cycle of expression signal pulse;

The employing high specific merges, and the multipath amalgamation result that obtains RAKE receiver is: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}---\left(3\right)$

Wherein, $y_i^{\left(m\right)\left(k\right)}=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}^{*}{y}_i^{\left(m\right)\left(k\right)}\left(l\right)---\left(4\right)$

A _IlBe Estimated value, a _IlThe channel fading value of representing i user l footpath, P _iThe power of expression user i, L are the footpath number;

Court verdict to m bit of i user is: ${\hat{a}}_i^{\left(m\right)\left(k\right)}=\mathrm{sgn}\left\{Y_i^{\left(m\right)\left(k\right)}\right\}---\left(5\right)$

(b.) the every bit decision result's of calculating reliability

In the k level PIC structure, the court verdict of m the bit of user i is , formula (3) can further be expressed as: $Y_i^{\left(m\right)\left(k\right)}=\mathrm{Re}\left\{y_i^{\left(m\right)\left(k\right)}\right\}={\mathrm{\μ}}_i{a}_i^{\left(m\right)}+n_i---\left(10\right)$

n _iBe white Gaussian noise, Normal Distribution

Be m the bit of user i, be worth and be+1 or-1, μ _iBe the real number relevant with channel fading,

Coefficient of reliability be calculated as follows: $f_i^{\left(m\right)\left(k\right)}=\tanh \left\{w\frac{{\mathrm{\μ}}_i\left|Y_i^{\left(m\right)\left(k\right)}\right|}{{\mathrm{\σ}}_i^2}\right\}---\left(15\right)$

Promptly be Coefficient of reliability, w is an arithmetic number, is used for remedying inaccurate that noise power estimates, when signal to noise ratio is higher, can get w=1;

(c.) the bit-level weighting of subscriber signal regeneration

According to the bit-level weighting regenerated signal of formula (14) calculating user i, its expression formula is: $g_i^{\left(k\right)}\left(t\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{A}_{\mathrm{il}}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{f}_i^{\left(n\right)\left(k\right)}{\hat{a}}_i^{\left(n\right)\left(k\right)}p(t-n{T}_b-{\mathrm{\τ}}_{\mathrm{il}})c_i(t-{\mathrm{\τ}}_{\mathrm{il}})---\left(14\right)$

Wherein, p (t) indication cycle is T _bSignal pulse;

(d.) calculating of multiple access interference

In the k level PIC algorithm, the multiple access of user i disturbs (MAI) to be: ${\hat{I}}_i^{\left(k\right)}=\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{g}_j^{\left(k\right)}\left(t\right)---\left(16\right)$

(e.) interference cancellation

If the weights of described double-weighing PIC algorithm k level are p ^(k), according to the following equation the MAI that obtains in the steps d is weighted interference cancellation, obtain the output signal of user i in the k level PIC structure Be the input signal of the RAKE receiver of user i in the next stage PIC structure, $r_i^{(k+1)}\left(t\right)=r\left(t\right)-p^{\left(k\right)}{\hat{I}}_i^{\left(k\right)}---\left(9\right)$

(f.) repeating step (a. ~ e.), carry out the calculating of next stage PIC, for afterbody PIC structure, only carry out the calculating of multi path despreading among the step a, multipath merging, according to described formula (2) input signal of user i is carried out multi path despreading, carry out multipath according to described formula (3) and described formula (4) and merge, multipath is merged the final result of the soft output of the user i that obtains as user i in the multistage PIC structure, in receiver, this result is fed to the decoder of user i and deciphers.

Images